Dear all,
I have few quick questions about what is behind the Johansen's VEC model that I hope someone could answer. I tried the manual and the web without luck. First, what optimization method does Eviews use for the in-built VEC object? And is the method different for estimating the cointegration vector from estimating the restricted VAR (or the VEC)? Second, does anyone have or know of a written program for the Johansen's VEC for Eviews? All these questions are because I would like to estimate a VEC step by step so to speak rather than use the "menu". Any suggestions are much appreciated!
Many thanks, c.
Johansen's VEC optimization method
Moderators: EViews Gareth, EViews Moderator
Re: Johansen's VEC optimization method
Eviews moderators, can you help at least with part of my question(s)?
-
EViews Glenn
- EViews Developer
- Posts: 2682
- Joined: Wed Oct 15, 2008 9:17 am
Re: Johansen's VEC optimization method
The optimization approachs used by EViews is as described in detail in the Lutkepohl book that we cite. As noted in that reference the estimation method will vary depending on the specification, but the general principal is ML. For example, the affine restriction case uses the method described by Boswijk 1995.
Unfortunately, I don't know of any EViews program that estimates a VEC.
Unfortunately, I don't know of any EViews program that estimates a VEC.
Re: Johansen's VEC optimization method
You can choose optimization method as you want provided that you set up your model appropriately.
In the attached workfile, two exemples are provided. First, the EViews method (which I guess is the least squares/eigenvalue method described in Lutkepohl or Juselius (see chapter 7 in "The cointegrated VAR model" for the step-by-step description)). The object in the workfile that is related to this calculation is a VAR-object.
But, as a second example, there is a SYSTEM-object. Provided that you type in your model correctly, you can estimate it by ML and choose whatever optimization method available to (hopefully) get the same results.
In the attached workfile, two exemples are provided. First, the EViews method (which I guess is the least squares/eigenvalue method described in Lutkepohl or Juselius (see chapter 7 in "The cointegrated VAR model" for the step-by-step description)). The object in the workfile that is related to this calculation is a VAR-object.
But, as a second example, there is a SYSTEM-object. Provided that you type in your model correctly, you can estimate it by ML and choose whatever optimization method available to (hopefully) get the same results.
- Attachments
-
- CI_test.WF1
- (18.84 KiB) Downloaded 429 times
Who is online
Users browsing this forum: No registered users and 2 guests
